Semantic Web-Based Information Systems: State-of-the-Art ...

Specific Reasoning as Theories
Query ng the Web Recons dered 0
Many practical applications require special forms of reasoning; for instance, efficient
equality reasoning is often performed using the so-called paramodulation
rule instead of the equality axioms (transitivity, substitution, and symmetry). Also,
temporal data might require conversions between different time zones and/or calendar
systems that are expressed in a simpler format and more efficiently performed using
arithmetic instead of logical axioms. Finally, reasoning with intervals of possible
values instead of exact values (e.g., for appointment scheduling) is conveniently
expressed and efficiently performed with constraint programming.
For this reason, it is desirable that a query language for the (conventional and
Semantic) Web can be extended with so-called “theories” implementing specific
forms of reasoning.
Such “theory extensions” can be realized in two manners:
• First, a theory can be implemented as an extension of the runtime system of the
query language with additional language constructs for using the extension.
• Second, a theory can be implemented using the query language itself and made
available to users of this query language through program libraries. In this
case, theories are implemented by rules and queries. Based upon the XML
syntax of the query language (Section 2.12), for example, such rule bases can
then be queried using the query language itself and maintained and updated
by a reactive language such as XChange (Bry & Ptrânjan, 2005).
Querying Ontologies and Ontology-Aware Querying
In a Semantic Web context, ontologies can be used in several alternative ways. First,
they can be dealt with by a specialized ontology reasoner (the main disadvantage
being the impossibility of adding new domain-specific constructs). Second, they can
be regarded as descriptions to be used by a set of rules implementing the Semantics
of the constructs employed by the ontology. (This is similar to a meta-interpreter and
may be slow.) Alternatively, the ontology may be “compiled” to a set of rules.
As discussed in the previous point, the query language should allow for both approaches:
extending the query language by specific theory reasoners for a certain
ontology language (e.g., OWL-DL) as well as the ability to use rules written in the
query language as means for implementing (at least certain aspects) of an ontology
language. Examples for such aspects are the transitivity of the subsumption
hierarchy represented in many ontologies or the type inference based on domain
and range restrictions of properties.
Copyright © 2007, Idea Group Inc. Copying or distributing in print or electronic forms without written permission
of Idea Group Inc. is prohibited.